Landscape change and the dynamics of openformations in a natural reserve

Landscape and Urban Planning 77 (2006) 167–177

Landscape change and the dynamics of openformations in a natural reserve

Duccio Rocchini a,∗, George L.W. Perry b, Mariangela Salerno a,Simona Maccherini a, Alessandro Chiarucci a

a Dipartimento di Scienze Ambientali “G. Sarfatti”, Universita di Siena, via P.A. Mattioli 4, 53100 Siena, Italyb School of Geography and Environmental Science, University of Auckland, Private Bag 92019, Auckland, New Zealand

Received 1 May 2004; received in revised form 24 February 2005; accepted 24 February 2005Available online 11 July 2005

Abstract

Remote sensing, when used in conjunction with landscape pattern metrics, is a powerful method for the study of ecologicaldynamics at the landscape scale by means of multi-temporal analysis. In this paper, we examine temporal change in openformations in the natural reserve of Poggio all’Olmo (central Italy). This area has undergone rural depopulation and the cessationof traditional methods of agriculture, resulting in the subsequent re-establishment and spread of other vegetation formations.Aerial photographs taken in 1954, 1977 and 1998 were orthorectified and classified based on the physiognomic characteristicsof the vegetation. An objective definition of the minimum mapping unit (MMU) was guaranteed by using vector format gridsfor this classification. We applied landscape pattern metrics based on landscape composition, the shape and size of patchesaflwi©

K

1

l

f

0

nd patch isolation. Our results demonstrate the key roles of shrubland, woodland and coniferous plantations in the ongoingragmentation of open formations in the landscape. Multi-temporal landscape analyses, and, in particular, a restricted suite ofandscape metrics, proved useful for detecting and quantitatively characterizing dynamic ecological processes. We concludeith some recommendations on the management alternatives feasible for the protection of the remaining grassland formations

n the natural reserve of Poggio all’Olmo.2005 Published by Elsevier B.V.

eywords: Aerial photos; Biodiversity management; GIS; Landscape metrics; Minimum mapping unit; Multi-temporal analysis

. Introduction

In many ecosystems, biodiversity is rapidly beingost, and Reid’s (1992) estimate of a 13% loss over

∗ Corresponding author. Tel.: +39 057 723 5408;ax: +39 057 723 2860.

E-mail address: [email protected] (D. Rocchini).

the period 1990–2015 is beginning to appear conser-vative (Nagendra and Gadgil, 1999). At the same time,new integrated approaches for ecosystem monitoringand management have proven increasingly success-ful in regional planning and natural resource man-agement (Innes and Koch, 1998). Landscape analysesconducted over large spatial extents are important forbiodiversity conservation (Forman and Godron, 1986;

169-2046/$ – see front matter © 2005 Published by Elsevier B.V.doi:10.1016/j.landurbplan.2005.02.008

168 D. Rocchini et al. / Landscape and Urban Planning 77 (2006) 167–177

Forman, 1995). The sheer complexity of many ecosys-tems suggests that characterizing ecosystems by meansof the more conventional species-level approach maybe either expensive or even impossible when histor-ical data are not readily available (Innes and Koch,1998; Roy and Tomar, 2000). Remote sensing andGeographical Information Systems (GIS), when inte-grated with the tools of landscape ecology, can beused to investigate the changing spatial patterns ofbiodiversity (Innes and Koch, 1998; Roy and Tomar,2000).

Of the features of the ecosystem most easily iden-tified via remote sensing, the physiognomy and spatialpattern of the vegetation are among the most important,since they characterize the landscape both structurallyand functionally. A description of the shape, size andspatial arrangement of patches of vegetation in the land-scape can be used to link the observed pattern with theecological processes that may have generated it. Forexample, the fragmentation of ecological units (a spa-tial characteristic) has often been reported to be relatedto ecosystem degradation (Ludeke et al., 1990; Mertensand Lambin, 1997).

Alongside spatial scale, temporal scale is equallyimportant when assessing the change of a landscapeover time. Multi-temporal analysis based on remotelysensed data (e.g. grey-scaled aerial photographs) hasplayed an important role in landscape ecology (Butayeet al., 1999; Defries and Townshend, 1999; Viedmaand Melia, 1999). In most studies, the classificationobppijr1ltrw(nfpaw

metrics. This method of interpretation, although verytime consuming, ensures that the analysis is undertakenusing a precise and consistent MMU, and thus allowsfor the comparison of different processes at the sameresolution.

The aim of this paper is to quantify landscape changein an abandoned rural landscape and give some recom-mendations on the management alternatives feasiblefor the protection of the grassland formations.

2. Study area

The natural reserve of Poggio all’Olmo, on theslopes of Mt. Amiata (latitude 42◦52′00′′, longitude11◦28′21′′, datum WGS84), occupies an area of about440 ha, with calcarenite, clay and sandstone being themain soil types. The elevation ranges from 664 to1016 m, and the climate belongs to humid type B2,at the limit with B1, in terms of Thornthwaite’s (1948)classification. Total annual rainfall averages 1045 mmand mean annual temperature averages 12.5 ◦C (cli-matic station: Castel del Piano, 639 m a.s.l.). Recently,the area has undergone a rapid re-colonization bywoody vegetation in open formations (pastures andabandoned fields). Land-use has changed considerablyin the area surrounding the reserve, particularly sincethe Second World War, due to rural depopulation and acessation/decrease in traditional agricultural practices(e.g. terracing, grazing, maintenance of grazed grass-lcall

vftaswbcPpto

f airborne data is performed by digitizing polygonsased on photo-interpretation, with the minimum map-ing unit (MMU) defined by means of the smallestolygon identified. This method leads to problemsn comparing maps whose MMU is defined so sub-ectively. Landscape metrics show variations across aange of scales (Turner et al., 1989; Jelinski and Wu,996; Wu et al., 1997, 2000, 2002), and so the trans-ation of information across scales is problematic ifhe MMU is not clearly defined. Moreover, all of theesampling methods commonly used to compare mapsith different input MMUs introduce some spatial error

Turner et al., 1989; Gustafson, 1998), since they doot take account of the ‘true’ resolution of the maps;or example, a map with a MMU of 100 m, but resam-led at 20 m MMU, will not reveal patterns recognizedt <100 m. In this paper, we used vector format gridsith a pre-defined cell size, alongside landscape pattern

ands by burning, cultivation of chestnut groves, wood-utting, cultivation of fields, maintenance of hedgerowst field boundaries). The natural reserve was estab-ished in 1998 to conserve remnant semi-natural grass-and habitats.

As described by Angiolini et al. (1999), theegetation found in the reserve today comprises: openormations, e.g. semi-natural grasslands (Bromus erec-us grasslands and Arrhenatherum elatius grasslands)nd small cultivated fields; Prunus spinosa and Cytisuscoparius shrublands; woodlands, e.g. Quercus cerrisoodlands, mixed broad-leaf woodlands (dominatedy Quercus pubescens, Ostrya carpinifolia and Q.erris), riparian woody vegetation (dominated byopulus nigra, Ulmus minor, Acer campestre, Tilialatyphyllos and O. carpinifolia or Salix sp. pl.), maplehickets and chestnut woodlands. Cultivated fieldsnce covered about one-third of the area (Comune

D. Rocchini et al. / Landscape and Urban Planning 77 (2006) 167–177 169

di Cinigiano, unpublished data) but are now almostcompletely absent. Exotic conifers (Abies alba Miller,Cupressus arizonica Green, Pinus nigra Arnold, Pinussylvestris L., Pseudotsuga menziesii (Mirbel) Franco),which do not reproduce spontaneously, were plantedin the 1950s–1970s, with two large plantations atPoggio la Torretta and Poggio Matoraio (Maccheriniet al., 2001). Current agro-forestry practices are largelyextensive, consisting of sheep grazing, wood-cutting(coppice with standards), cultivation of chestnutgroves and small arable fields.

3. Methods

3.1. Derivation of land-use maps

Aerial photographs (grey-scale) taken in 1954,1977 and 1998 (scale: 1:33,000) were scanned athigh resolution (600 dpi). Orthorectification, basedon a Digital Terrain Model (DTM) derived froma 1:10000 topographic map (pixel size: 10 m) ofthe study area and on 30 ground control points(GCPs), was performed using ERDAS IMAGINE 8.4(www.erdas.com). Each image’s spatial resolution wasapproximately 2 m. Positional accuracy was tested bymeans of 20 additional GCPs; error never exceeded4 m. Images were projected onto the National (Italian)Coordinate System (Gauss Boaga Projection, datumRoma 40).

ow4Atattmgecas(tt

3.2. Landscape analysis

For the landscape analysis, vector format grids wereconverted into a raster format at a cell resolution of10 m × 10 m, and a suite of non-redundant landscapepattern metrics was calculated for each of the grids.Landscape composition was quantified by means ofthe area covered by each class. Transition matrices,based on cells occupied by classes over the three timeperiods, were generated to quantify the overall gainsand losses of area (converted from cells) for the differ-ent landscape classes. The Kappa (K) statistic (Chustet al., 1999; Romero-Calcerrada and Perry, 2004) wasused to quantify how stable the different vegetationclasses were: this index is based on a comparisonbetween pairs of maps and ranges from 0 (no pixelschange) to 1 (all pixels change). We used a series oflandscape pattern metrics to quantify change in thelandscape. Such metrics have been widely applied insimilar studies, and allow objective description of thetemporal pattern of landscape change and comparisonof this with other (similar) landscapes (Turner et al.,2001). Landscape structure was assessed by means ofpatch-based metrics, such as the total and per classnumber of patches, shape-based metrics, such as theArea Weighted Mean Shape Index (AWMSI) and size-based metrics, such as the maximum, minimum, meanand standard deviation of patch size (MaxPS, MinPS,MPS, PSSD, respectively). Shape metrics approachzero when patches have perfectly square (raster for-mmyssstlrsTpbprw11

The minimum mapping unit was defined a pri-ri by the superimposition of a vector format gridith a cell size of 10 m × 10 m, giving a total of4,000 cells; this analysis was conducted using thercView 3.2 GIS (ESRI). Each cell was subjected

o photo-interpretation by means of pixel radiancend the physiognomic characteristics of the vegeta-ion it contained. Vegetation classes recognized usinghese methods were: woodlands, shrublands, open for-

ations (pastures and fields), buildings, isolated orrouped trees, linear formations (hedges) and conif-rous plantations. If a cell contained two or morelasses, the value of the prevalent class (in terms ofrea) was used. This procedure was performed on theame 10 m × 10 m grid for each of the three periods1954, 1977, 1998); thus, each cell represents a spa-ial unit whose identity may, potentially, change overime.

at) or circular (vector format) shapes. For the sizeetrics, frequency–size distributions were also anal-

sed; these allow an analysis of changes in the patchtructure through time (e.g. what proportion of the land-cape is covered by patches below a certain size). Thelope of the patch size versus frequency plot quantifieshe ratio of the number of large to small patches in theandscape; as the slope increases, large patches becomearer with respect to small ones. Isolation was mea-ured by means of the Mean Proximity Index (MPI).his index is directly related to the area covered byatches and inversely related to the nearest neigh-our distance, and represents the “non-isolation” ofatches. The MPI equals zero if all patches of the cor-esponding class have no neighbour of the same typeithin a given search radius (McGarigal and Marks,995); in these analyses the search radius was set to00 m. A thorough explanation of these metrics is given

http://www.erdas.com/


by McGarigal and Marks (1995) and Turner et al.(2001).

4. Results

4.1. General temporal trends

The three maps show that over the last 50 years im-portant changes have occurred in the landscape (Fig. 1).

In 1954, open formations were the most abundant anddominant class, and they constituted a matrix withinwhich patches of the other vegetation classes weredispersed. By 1977, a considerable decrease in theabundance of open formations, with commensurateincreases in the other classes, had taken place. Thisresulted in increased fragmentation; a clearly dominantmatrix was no longer recognizable and the landscapemosaic showed a higher heterogeneity, in terms of spa-tial structure. By 1998, open formations had decreased

Fig. 1. Change in vegetation cover over the period 1954–199
8 derived from interpretation of aerial photographs.


Fig. 2. Area (as a percentage of total reserve area) occupied by eachclass. White columns: 1954; grey columns: 1977; black columns:1998.

still further and woody vegetation (woodlands andconiferous plantations) had increased.

4.2. Landscape composition

The total area of woodland increased from 28% in1954 to 34% in 1977 and to 41% in 1998 (Fig. 2). Therelative increase in woodland abundance, from 1954to 1998, was 48%. Coniferous plantations were notpresent in the landscape in 1954, but constituted 15% ofthe total area in 1998. Shrublands represented the vege-tation type with the highest relative increase, being fivetimes more abundant in 1998 than in 1954, and coveringabout a quarter of the total area. Conversely, open for-mations underwent a consistent decrease; cover of thisclass declined from 65% in 1954 to 19% in 1998, with aloss of more than 70% of the area covered in 1954. By1998, this class was largely replaced by shrublands,woodlands and coniferous plantations. Isolated treesand linear formations (hedges), although always onlypresent at low abundances, also decreased markedly;

Fig. 3. Kappa values for each class, representing the level of perma-nence. White columns: 1954–1977 period; grey columns: 1977–1998period; black columns: 1954–1998 period. Note that the only non-zero value for the ‘coniferous plantations’ class is for the interval1977–1998; this is because this vegetation type was absent from thelandscape in 1954 making comparison meaningless.

both of these vegetation classes suffered about a 70%decline in cover over the period 1954–1998.

Based on transitions occurring in the period from1954 to 1977, woodlands and shrublands were foundto be highly stable (Table 1). Conversely, open for-mations were frequently transformed into woodlands,shrublands or coniferous plantations. Isolated trees andlinear formations showed low stability, losing 78.50and 73.85% of their areal cover, respectively. A sim-ilar pattern was observed in the period from 1977 to1998 (Table 2), with some differences resulting fromthe presence of coniferous plantations, which increas-ingly replaced the open formations, isolated trees andlinear formations. The Kappa statistic (Fig. 3) indicatedthat woodlands, shrublands, coniferous plantations andbuildings were the most stable components of the land-

Table 1Transition matrix for 1954–1977 period (values in hectares)

1954 1977

Woodlands Shrublands Open formations Buildings Isolated orgrouped trees

Linearformations

Coniferousplantations

Woodlands 91.5 13.3 11.2 0 0.2 0.2 6.7Shrublands 1.4 11.8 4 0 0 0.2 1.1O 0 1.5 0.6 36.9B 0.2 0 0 0I 0 0.4 0 0.1L 0 0.1 3.4 1.8

pen formations 55.7 57.7 132.2uildings 0 0 0

solated or grouped trees 0.9 0.2 0.4inear formations 0.7 3.8 2.8


Table 2Transition matrix for 1977–1998 period (values in hectares)

1977 1998

Woodlands Shrublands Openformations

Buildings Isolated orgrouped trees

Linearformations

Coniferousplantations

Woodlands 134.7 2.8 2.1 0 0 0.1 10.3Shrublands 19 56.1 9.2 0 0.1 0.9 1.4Open formations 24.1 40.6 70.5 0 0.2 1.1 13.9Buildings 0 0 0 0.3 0 0 0Isolated or grouped trees 0.2 0.7 0.7 0 0.3 0 0.3Linear formations 0.8 1.1 0.1 0 0 0.1 0.6Coniferous plantations 3.8 0.2 1.4 0 0 0 41.2

scape, while open formations, isolated or grouped treesand linear formations were the least stable. This is con-sistent with the patterns of change in the landscapesince 1954 described above.

4.3. Landscape structure

The total number of patches in the landscapeincreased by about 20% from 1954 to 1977 (609–734)and then declined by about 36.6% from 1977 to 1998(734–465), resulting in an overall decrease of 23.6% forthe entire period (1954–1998). In terms of the numberof patches per class (Fig. 4), there was a progressivedecrease from 1954 to 1998 in all classes, with theexception of shrublands and open formations, whichshowed an increase from 1954 to 1977 and a subse-quent decrease from 1977 to 1998. The number of shrubpatches in 1998 was half that of the original number,while the number of patches of open formations wassimilar to that seen in 1954.

In terms of patch shape, the open formations showedthe highest value of AWMSI in 1954, but there was

Fc

an abrupt decrease in the values of this index overtime (Fig. 5). The shape of woodland patches becameincreasingly simple. By contrast, shrublands showeda trend towards having more diffused and complexshapes. No discernible trends were observed in theshape of patches of buildings, isolated trees, linear for-mations and coniferous plantations.

Over the three “sampling dates”, PSSD values werehigher than MPS values, for all classes (Fig. 6), sug-gesting a right-skewed distribution with small patchesbeing numerically dominant for all classes (see alsoFig. 8). This may be related to the classification method,using a fixed grid cell size at a very fine resolution. Allclasses showed an increase in MPS over time, with theexclusion of open formations.

In 1954, the minimum patch size for all classes was0.01 ha (the MMU), with the exception of coniferousplantations (0.08 ha). By contrast, the MinPS valuesin 1998 reveal the aggregated and artificial nature of

Fctf

ig. 4. Number of patches per class. White columns: 1954; greyolumns: 1977; black columns: 1998.

ig. 5. Area Weighted Mean Shape Index (AWMSI) per class. Whiteolumns: 1954; grey columns: 1977; black columns: 1998. Noticehe high value of open formations in 1954 and the abrupt decrease inollowing dates, showing a move towards a simpler shape over time.


Fig. 6. Patch size: mean patch size (MPS), columns and patch sizestandard deviation (PSSD), error bars, per class. White columns:1954; grey columns: 1977; black columns: 1998.

the coniferous plantations. As for the MPS (Fig. 7),open formations had the highest value in 1954, but theysuffered a remarkable decrease over subsequent years.Woody vegetation (woodlands, shrublands, coniferousplantations) showed opposite trends to the open forma-tions. The frequency–size distribution of patch areas(Fig. 8) showed large differences between the maps of1954, 1977 and 1998, with a steeper slope in 1954 cor-responding to the abundance of small-medium patches,while a shallower slope was found in 1977 and 1998.

The value of the MPI for open formations in 1954exceeded that of all other patch types (Fig. 9). Openformations showed an abrupt decrease from 1954 to1977, and a further decrease from 1977 to 1998, sug-gesting their increased fragmentation over this period.Shrublands and coniferous plantations showed a pro-gressive increase in the MPI, suggesting a decrease intheir isolation. Woodlands showed a different trend: alack of change from 1954 to 1977 and then a decrease

F1

Fig. 8. Patch size frequency–area distributions for each year; lightgrey squares: 1954; dark grey circles: 1977; black triangles: 1998.

Fig. 9. Mean Proximity Index (MPI) per class. White columns: 1954;grey columns: 1977; black columns: 1998.

from 1977 to 1998. Buildings, isolated trees and lin-ear formations had the lowest values of MPI, resultingfrom their scattered occurrence in the landscape.

5. Discussion

The cessation/decrease of traditional agriculturalpractices in the natural reserve of Poggio all’Olmohas resulted in a decline in the abundance of openformations, linear formations (hedges) and isolated orgrouped trees, and the spread of woody vegetation(shrublands, woodlands, coniferous plantations). Theopen formations have become increasingly fragmentedsince 1954, when a single patch occupied about 60% ofthe total area; by 1977 and then 1998 these formationshad become fragmented into several small patches witha commensurate increase in patch isolation. The spreadin shrubland areas largely accounts for this increase

ig. 7. Maximum patch size per class (MaxPS). White columns:954; grey columns: 1977; black columns: 1998.


in landscape fragmentation. This vegetation type hasexpanded by forming a number of patches character-ized by diffuse edges. Over time these patches havecoalesced and “closed” the previously open formations,linear formations (hedges) and isolated or groupedtrees. This phenomenon is basically related to the estab-lishment of several nuclei shrubs, which initially colo-nize the open areas in a stochastic manner, and becomecloser to one another over time, before being replacedby woodland patches. This invasion of the open for-mations by isolated woody shrubs, and the subsequentcoalescence of these woody invasive species into largerpatches explains the trend in the number and size ofpatches of open formation vegetation—an increase dur-ing the early invasion phase followed by a decrease asindividual patches coalesce. Other authors (e.g. Loehleet al., 1996) have speculated that, once the woodyvegetation cover exceeds a critical threshold, this col-onization process will ‘self-accelerate’. Woody forma-tions (woodlands and coniferous plantations) increasedfrom less than 25% to about 50% of the reserve area(1954–1998), whereas the area of open formations hasfallen by 70%, when compared to the area occupiedin 1954. These values emphasize the magnitude ofchanges that have led to, and will continue to resultin, the replacement and loss of the open formations.

The expansion of woody vegetation over open veg-etation types, following the cessation of traditionalmanagement, has been observed in several Mediter-ranean areas (e.g. see Blondel and Aronson, 1999;CPaTlpetmoeaidepio

a commensurate decrease in the cover of herbaceousvegetation, from 56% in 1964 to 24% in 1998. Theirresults show that the regeneration of woodland vegeta-tion can occur quite rapidly (certainly over a period of30 years).

In Italy, similar trends have been observed in variousparts of the Apennine mountains, such as the SolanoValley (Vos and Stortelder, 1992), the Aterno Valley(Peroni et al., 2000), the Aniene Middle Valley (Blasiet al., 2001), the Northern Apennines (Torta, 2004) andthe Pisan Hills (Bertacchi and Onnis, 2004). Blasi andDi Pietro (1998) and Di Pietro and Filibeck (2000)report that 30–40 years of abandonment were suffi-cient for the regeneration of woodland coenoses in hillyareas characterized by terracing in Central and South-ern Italy. Other authors, however, have found that itcan take much longer for woodland colonization ofabandoned unterraced agricultural fields in Mediter-ranean areas to occur (e.g. Escarre et al., 1983). AsDebussche and Lepart (1992) conclude, it is difficult tomake broad generalisations about the invasion of oldfield sites by woody vegetation in the Meditteranean.Nevertheless, Vos and Stortelder (1992) consider thatthe “remnants” of traditional Tuscan mountain land-scape will have changed radically in another 50 yearsand may totally vanish. These concerns are echoed byBertacchi and Onnis (2004), who note that modifica-tion of the ‘typical’ landscape in Tuscany is prevalentin hilly areas close to urban centres and on foothills atthe edge of plains. In the present study, a 50-year timeplosllispadst(aaig

armel and Kadmon, 1999; Romero-Calcerrada anderry, 2004), including elsewhere in central Italy (Vosnd Stortelder, 1992; Peroni et al., 2000; Torta, 2004).he general successional trend in many Mediterranean

andscapes is from pasture to shrublands (abandonedastures) and woody vegetation; the latter is consid-red to be the endpoint of secondary succession inhese ecosystems. This type of old-field succession

ay have important implications for a wide rangef ecological processes such as, for example, nutri-nt cycling (Blondel and Aronson, 1999; Wainwrightnd Thornes, 2004). Carmel and Kadmon (1999) stud-ed the long-term effect of grazing on the vegetationynamics in a Mediterranean ecosystem in the North-rn Galilee Mountains, Israel, from 1964 to 1998. Theyoint out that a decrease in grazing intensity resultedn an increased cover of woodland coenoses, from 2%f total area in 1964 to 41% in 1998, accompanied by

eriod has been sufficient for the regeneration of wood-and coenoses. The observed decrease in the numberf woodland patches corresponds with simpler patchhapes developing as individual patches begin to coa-esce. Likewise, previously separate patches of wood-ands and coniferous plantations have merged, resultingn the formation of fewer and larger patches, again withimpler shapes. For coniferous plantations, this is sim-ly a consequence of the way in which such plantationsre managed. As for woodlands, this pattern of forestynamics has been already recognized at other Italianites that have undergone abandonment and a cessa-ion of traditional forestry and agricultural practicesVos and Stortelder, 1992; Peroni et al., 2000; Blasi etl., 2001). As Vos and Stortelder (1992) and Bertacchind Onnis (2004) discussed, the fine-grained pattern-ng seen in some traditional agricultural landscapes isradually being lost and is progressively replaced by


a more coarsely grained pattern, characterized by thedominance of woodland assemblages over the othervegetation types. In this study, a significant decrease inthe total number of patches was observed (1977–1998),after an initial increase in the period 1954–1977.From 1977, the change in landscape structure, andthe subsequent fragmentation of open formations, hasresulted in the isolation of patches belonging to thisclass. These patches have become embedded withinthe woody classes (shrublands, woodlands, conifer-ous plantations). The reduced patchiness observed in1998 fits the conceptual model of ‘patch fusion’ forthe dominant vegetation type (woodlands, see Vos andStortelder, 1992), which has replaced the other vegeta-tion types following a change in land-use management.

6. Implications for restoration management

The changes in the open formations that occurredin the natural reserve of Poggio all’Olmo are typicalof a general trend occurring across Europe; in partic-ular, semi-natural grassland communities have, due tochanges in land-use, declined in area and quality dur-ing the second half of the 20th century across all ofEurope (van Dijk, 1991; Bakker and Berendse, 1999).These plant communities are of great interest, due totheir landscape aesthetic value, the high number ofrare and/or endangered species they harbour, and theirhigh species richness (see Willems, 2001). Abandonedpfgt(

eaugao

oBioe

Bissels et al., 2004). This slows restoration, especiallyfor formerly arable land, if target species are not sownand the establishment rate of competitive, early suc-cessional species, is not artificially reduced (Hutchingsand Booth, 1996; Kleijn, 2003).

The restoration of grassland vegetation after coniferforestry also seems to be a slow and difficult process(Csontos et al., 1996, 1996/1997; Maccherini and DeDominicis, 2003), as the grasslands are overtoppedby trees (Dzwonko and Loster, 1998). Encouragingresults, however, have been obtained in the restora-tion of grassland overgrown by shrubs, by the clear-ing of shrubs and mowing or grazing (Bobbink andWillems, 1993; Zobel et al., 1996; Willems and Bik,1998; Barbaro et al., 2001). An additional problem forthe management of protected areas is the presence ofexotic species (D’Antonio and Meyerson, 2002): theycan out-compete native species, infect them with dis-eases, out-compete them, or otherwise alter ecosystemfunctioning, making making the restoration of the orig-inal ecosystem a difficult and expensive task (Vitouseket al., 1997).

Based on these considerations, restoration of semi-natural grasslands for the natural reserve of Pog-gio all’Olmo is recommended only for the areasencroached by shrubs, on calcarenite by appropriatemanagement (cutting of shrubs, burning and grazing);the same actions are also recommended for mainte-nance of the remaining semi-natural grasslands in therest of the natural reserve. Abandoned ex-arable landsosrctmcssah

R

A

astures and lowered grazing pressure are the mainactors triggering natural succession of semi-naturalrasslands into increasingly dense shrublands, charac-erized by lower species richness and fewer rare speciesBarbaro et al., 2001).

Pastures are the vegetation type containing the high-st number of species of phytogeographic interestnd are the most threatened habitat within the nat-ral reserve (Maccherini et al., 2001); semi-naturalrasslands on calcareous substrata are also classifieds habitats of ‘community importance’ in Annexure If the Habitats Directive 92/43/ECC (Romao, 1996).

The restoration of semi-natural grasslands dependsn the degree of degradation (Muller et al., 1998;akker and Berendse, 1999). An important factor lim-

ting the success of this type of management is thatften only a few grassland species remain present inither the (soil) seed bank and/or the seed rain (see

n clay and sandstone are actually covered by densehrublands and are very hard to restore. Given theireduced value as habitat of rare and phytogeography-ally interesting species they can optimally be left toheir natural succession processes. On the contrary,

aintenance of hedgerows, riparian woody vegetation,hestnut groves, maple thickets and the gradual sub-titution of exotic conifers with native woody specieshould be considered a priority for conservation ofhigh ecosystem heterogeneity and, consequently, a

igh species diversity.

eferences

ngiolini, C., Maccherini, S., Chiarucci, A., Gabellini, A., DeDominics, V., 1999. Memoria illustrativa alla carta della veg-etazione della Riserva Naturale “Poggio all’Olmo” (Grosseto,Toscana Meridionale). Atti. Mus. Stor. Nat. Maremma 19, 29–47.


Bakker, J.P., Berendse, F., 1999. Constraints in the restoration of eco-logical diversity in grassland and heathland communities. Tree14, 63–68.

Barbaro, L., Dutoit, T., Cozic, P., 2001. A six-year experimentalrestoration of biodiversity byshrub-clearing and grazing in cal-careous grasslands of French Prealps. Biodivers. Conserv. 10,119–135.

Bertacchi, A., Onnis, A., 2004. Changes in the forested agricul-tural landscape of the Pisan Hills (Tuscany, Italy). In: Maz-zoleni, S., di Pasquale, G., Mulligan, M., Martino, P., Rego,F. (Eds.), Recent Dynamics of the Mediterranean Vegetationand Landscape. Wiley & Sons Ltd., Chichester, U.K., pp. 167–178.

Bissels, S., Holzel, N., Donath, T.W., Otte, A., 2004. Evaluationof restoration success in alluvial grasslands under contrastingflooding regimes. Biol. Conserv. 118, 641–650.

Blasi, C., Di Pietro, R., 1998. Two new phytosociological types ofQuercus pubescen s. l. Woodland community in southern Latium.Plant Biosys. 132, 207–223.

Blasi, C., Fortini, P., Carranza, M.L., Ricotta, C., 2001. Analisi delladiversita del paesaggio vegetale e dei processi di recupero nellamedia valle dell’Aniene (Appennino centrale, Lazio). Fitosoci-ologia 38, 3–11.

Blondel, J., Aronson, J., 1999. Biology and Wildlife of the Mediter-ranean Region. Oxford University Press, Oxford, U.K.

Bobbink, R., Willems, J.H., 1993. Restoration management of aban-doned chalk grassland in the Netherlands. Biodivers. Conserv. 2,616–626.

Butaye, J., Honnay, O., Hermy, M., 1999. Vegetation mapping as anaid in detecting temporal vegetation changes in the Demer valley(Belgium). Belg. J. Bot. 132, 119–140.

Carmel, Y., Kadmon, R., 1999. Effects of grazing and topographyon long-term vegetation changes in a Mediterranean ecosystemin Israel. Plant Ecol. 145, 243–254.

Chust, G., Ducrot, D., Riera, J.L.L., Pretus, J.L.L., 1999. Charac-

C

C

D

D

D

D

Dzwonko, Z., Loster, S., 1998. Dynamics of species richness andcomposition in a limestone grassland restored after tree cutting.J. Veg. Sci. 9, 387–394.

Escarre, J., Houssard, C., Debussche, M., Lepart, J., 1983. Evolu-tion de la vegetation et du sol apres abandon cultural en regionmediterraneenne: etude de succession dans le Garrigues du Mont-pellierais (France). Acta Oecol. 4, 221–239.

Forman, R.T.T., 1995. Land Mosaic: the Ecology of Landscapes andRegions. Cambridge University Press, Cambridge.

Forman, R.T.T., Godron, M., 1986. Landscape Ecology. Wiley &Sons, Chichester/New York/Brisbane/Toronto/Singapore.

Gustafson, E.J., 1998. Quantifying landscape spatial pattern: what isthe state of the art? Ecosystems 1, 143–156.

Hutchings, M.J., Booth, K.D., 1996. Studies on the feasibility ofre-creating chalk grassland vegetation on ex-arable land. I. Thepotential roles of seed bank and seed rain. J. Appl. Ecol. 33,1171–1181.

Innes, J.L., Koch, B., 1998. Forest biodiversity and its assessment byremote sensing. Global Ecol. Biogeogr. Lett. 7, 397–419.

Jelinski, D.E., Wu, J., 1996. The modifiable areal unit problemand implications for landscape ecology. Landscape Ecol. 11,129–140.

Kleijn, D., 2003. Can establishment characteristics explain the poorcolonization success of late successional grassland species onex-arable land? Restor. Ecol. 11, 131–138.

Loehle, C., Li, B.L., Sundell, R.C., 1996. Forest spread and phasetransitions at forest-prairie ecotones in Kansas, USA. LandscapeEcol. 11, 225–235.

Ludeke, A.K., Maggio, R.C., Reid, L.M., 1990. An analysis ofanthropogenic deforestation using logistic regression and GIS.J. Environ. Manage. 31, 247–259.

Maccherini, S., Chiarucci, A., Selvi, F., De Dominicis, V., 2001. Floravascolare della Riserva Naturale di Poggio all’Olmo (Cinigiano,Grosseto). Atti. Soc. Tosc. Sci. Nat., Mem, Serie B 108, 27–41.

Maccherini, S., De Dominicis, V., 2003. Germinable soil seed-bank

M

M

M

N

P

R

R

terizing human-modelled landscapes at a stationary state: a casestudy of Minorca, Spain. Environ. Conserv. 26, 322–331.

sontos, P., Horanszky, A., Kalapos, T., Lokos, L., 1996. Seedbank of Pinus nigra plantations in dolomite rock grasslandhabitats, and its implications for restoring grassland vegetation.Annales Historico-Naturales Musei Nationalis Hungarici 88,69–77.

sontos, P., Tamas, J., Kalapos, T., 1996/1997. Soil seed banks andvegetation recovery on Dolomite hills in Hungary. Acta Bot.Hung. 40, 35–43.

’Antonio, C., Meyerson, L.A., 2002. Exotic plant species as prob-lems and solutions in ecological restoration: a synthesis. Restor.Ecol. 10, 703–713.

ebussche, M., Lepart, J., 1992. Establishment of woody plants inMediterranean old fields: opportunity in space and time. Land-scape Ecol. 6, 133–145.

efries, R.S., Townshend, J.R.G., 1999. Global land cover char-acterization from satellite data: from research to operationalimplementation? Global Ecol. Biogeogr. 8, 367–379.

i Pietro, R., Filibeck, G., 2000. Terrazzamenti abbandonati e recu-pero della vegetazione spontanea: il caso dei Monti Aurunci.Informatore Botanico Italiano 32, 17–30.

of former grassland converted to coniferous plantation. Ecol. Res.18, 739–751.

cGarigal, K., Marks, B.J., 1995. Fragstats: spatial pattern analysisprogram for quantifying landscape structure. USDA For. Serv.Gen. Tech. Re PNW-351.

ertens, B., Lambin, E.F., 1997. Spatial modelling of deforestationin southern Cameroon: spatial disaggregation of diverse defor-estation processes. Appl. Geogr. 17, 143–162.

uller, S., Dutoit, T., Alard, D., Grevilliot, F., 1998. Restoration andrehabilitation of species-rich grassland ecosystem in France: areview. Restor. Ecol. 6, 94–101.

agendra, H., Gadgil, M., 1999. Satellite imagery as a tool formonitoring species diversity: an assessment. J. Appl. Ecol. 36,388–397.

eroni, P., Ferri, F., Avena, G.C., 2000. Temporal and spatial changesin a mountainous area of central Italy. J. Veg. Sci. 11, 505–514.

eid, W.V., 1992. How many species will there be? In: Whitmore,T.C., Sayer, J.A. (Eds.), Tropical Deforestation and SpeciesExtinction. Chapman and Hall, London, pp. 55–73.

oy, P.S., Tomar, S., 2000. Biodiversity characterization at landscapelevel using geospatial modelling technique. Biol. Conserv. 95,95–109.


Romao, C. 1996. Interpretation manual of European Union habitats(Version EUR 15)—Edit. Directorate General XI “Environment,Nuclear Safety and Civil Protection” of the European Commis-sion, Bruxelles.

Romero-Calcerrada, R., Perry, G.L.W., 2004. The role of land aban-donment in landscape dynamics in the SPA Encinares del rıoAlberche y Cofio, Central Spain, 1984–1999. Landsc. UrbanPlan. 66, 217–232.

Thornthwaite, C.W., 1948. An approach towards a rational classifi-cation of climate. Geogr. Rev. 38, 55–94.

Torta, G., 2004. Consequences of rural abandonment in a north-ern Apeninnes landscape (Tuscany, Italy). In: Mazzoleni, S.,di Pasquale, G., Mulligan, M., Martino, P., Rego, F. (Eds.),Recent Dynamics of the Mediterranean Vegetation and Land-scape. Wiley & Sons Ltd., Chichester, U.K., pp. 167–178.

Turner, M.G., O’Neill, R.V., Gardner, R.H., Milne, B.T., 1989.Effects of changing spatial scale on the analysis of landscapepatterns. Landscape Ecol. 3, 153–162.

Turner, M.G., Gardner, R.H., O’Neill, R.V., 2001. Landscape Ecol-ogy in Theory and Practice: Pattern and Process. Springer-Verlag,New York.

van Dijk, G., 1991. The status of semi-natural grasslands in Europe.In: Goriup, P.D., Batten, L.A., Norton, J.A. (Eds.), The Conser-vation of Lowland Dry Grassland Birds in Europe. JNCC Publ.,Newbury, pp. 15–36.

Viedma, O., Melia, J., 1999. Monitoring temporal changes in the spa-tial patterns of a Mediterranean shrubland using Landsat images.Divers. Distrib. 5, 275–293.

Vitousek, P.M., D’Antonio, M., Loope, L.L., Rejmanek, M., West-brooks, R., 1997. Introduced species: a significant component ofhuman-caused global change. N. Z. J. Ecol. 21, 1–16.

Vos, W., Stortelder, A., 1992. Vanishing Tuscan Landscapes. PudocScientific Publishers, Wageningen, The Netherlands.

Wainwright, J., Thornes, J.B., 2004. Environmental Issues in theMediterranean: Process and Perspectives From the Past to the

W

W

Wu, J., Gao, W., Tueller, P.T., 1997. Effects of changing spatial scaleon the results of statistical analysis with landscape data: a casestudy. Geogr. Inf. Sci. 3, 30–41.

Wu, J., Jelinski, D.E., Luck, M., Tueller, P.T., 2000. Multiscale analy-sis of landscape heterogeneity: scale variance and pattern metrics.Geog. Inf. Sci. 6, 6–19.

Wu, J., Shen, W., Sun, W., Tueller, P.T., 2002. Empirical patterns ofthe effects of changing scale on landscape metrics. LandscapeEcol. 17, 761–782.

Zobel, M., Suurkask, M., Rosen, E., Partel, M., 1996. The dynam-ics of species richness in an experimentally restored calcareousgrassland. J. Veg. Sci. 7, 203–210.

Dr. Duccio Rocchini is a GIS analyst at the Department of Envi-ronmental Science, University of Siena, principally concerned withremote sensing approaches applied to biodiversity characterization.He teaches Thematic Cartography at the University of Siena.

Dr. George L.W. Perry (PhD, Melbourne) is a lecturer in the Schoolof Geography & Environmental Science at the University of Auck-land, New Zealand. He has interests in spatial and landscape ecologyand in particular the use of spatial modelling and analysis. He hasworked on modelling fire regime dynamics and landscape change inNew Caledonia, New Zealand and Spain. Main current field loca-tions are south-eastern (Victoria) and south-western (WA) Australia,and forest ecosystems in northern New Zealand.

Dr. Mariangela Salerno is a teacher of natural science in high privateschool; she has great interest about study of vegetation patterns byremote sensing.

Dr. Simona Maccherini is a permanent researcher at the Universityof Siena. Specific branches of her research interests are: the ecologyof semi-natural grasslands, the assessment of species diversity inprotected areas and the monitoring of restoration of semi-naturalh

Psabc

Present. Routledge, London, U.K.illems, J.H., 2001. Problems, approaches, and results in restoration

of Dutch calcareous grassland during the last 30 years. Restor.Ecol. 9, 147–154.

illems, J.H., Bik, L.P.M., 1998. Restoration of high species densityin calcareous grassland: the role of seed rain and soil seed bank.Appl. Veg. Sci. 1, 91–100.

abitat.

rof. Alessandro Chiarucci is associate professor at the Univer-ity of Siena with interests in community ecology and biodiversityssessment. The main focus of its research is on plant communities,ut with some looks to the species assemblages of other taxa. He isurrently an associate editor of the Journal of Vegetation Science.

Landscape change and the dynamics of openformations in a natural reserve

Documents

Transcript of Landscape change and the dynamics of openformations in a natural reserve